Post on 04-Jul-2020
FOOT AND ANKLE SAFETY EVALUATION IN REAL LIFE CRASH SITUATIONS
Niklas Höglund1, Per Lövsund1, David Viano1, 2, Stefan Olsén2
1. Chalmers University of Technology
2. Saab Automobile AB
ABSTRACT
The relation between crash severity and lower leg responses are studied using mathematical
models, not only with the 50th percentile Hybrid III dummy, but also using biofidelic feet model and a
female dummy model. A number of countermeasures, including toepan padding and foot airbag, were
evaluated in a broad spectrum of crash situations, occupant sizes and foot positions.
The simulations showed that an airbag in the foot area optimized for the most severe crash
reduced forces higher than 3.5 kN in the right tibia in most crash situations. However, tibial axial
forces were higher in 50% of the simulations with a foot bag optimized for a severe crash than without
foot air bag in various crash conditions. The biofidelic foot and ankle model rotated less than the
Hybrid III foot. The difference in tibia load was relatively small between the dummy model and the
human foot model.
The study shows that an implementation of a countermeasure in cars has to be preceded by an
investigation of dummy responses in several crash situations to avoid sub-optimization of the car
safety performance.
Keywords: Foot and ankle, mathematical modelling, parametric investigation, safety device
FOOT AND ANKLE INJURIES are not fatal but are still common in frontal car collisions
(Kuppa et al., 2001) and often result in disabling injuries (MacKenzie et al., 1993). It is therefore
considered necessary to reduce the frequency and the injury severity. Short drivers and females have
been proposed for higher risk for foot and ankle injuries than large drivers and males in
epidemiological studies by e.g. Dischinger et al. (1994) and Crandall et al. (1996).
The injury mechanism for the most severe injuries is a load in the axial direction of the tibia
(Manning et al., 1998). Critical axial tibial load is 5.2 kN and 3.75 kN for 50th percentile male and
5th percentile female respectively (Kuppa et al., 2001). A hyper dorsiflexion of the ankle joint also
influence injury outcome, especially talus and malleolar fractures (Begeman et al., 1993,
Manning et al., 1998).
Foot and ankle injuries are more likely to occur in crashes with large delta-v (Fildes et al., 1995,
Richter et al., 2001). Toepan intrusion and pedals have been associated with lower limb injuries
(Thomas et al., 1995, Crandall et al., 1998). A more throughout literature review is presented in
Höglund et al. (2003). Previous studies have shown that there is a complex relation between different
compartment deformation characteristics and lower leg responses (Höglund et al., 2003,
Höglund et al., 2002). A foot airbag has shown to be effective reducing tibia index and tibial axial
forces in a series of sled tests by Håland et al. (1998), Kippelt et al. (1998), and Hesse et al. (2001).
The Hybrid III dummy is the most commonly used crash test dummy, but the foot and ankle
joints lack in biofidelity. It is thus a need for a better understanding of the mechanisms causing lower
leg using more detailed models. The objective of this study is to evaluate the safety performance of a
car and countermeasures in situations similar to real life crashes.
METHODOLOGY
An occupant-vehicle model of a mid-size, left hand side drive, production car was developed and
compared to crash test data and sled test data. The toepan and carpet were then replaced by a toepan
from a conceptual model. The model with the conceptual toepan will be hereby being denominated as
IRCOBI Conference – Lisbon (Portugal), September 2003 163
the baseline model. The baseline model was used in parametric studies, divided in two parts;
evaluation of the safety performance of the car model and an evaluation of countermeasures
(Figure 1).
Definitions Study 1 Study 2a, parameters with fixed levels Study 2b, stochastic
levels of the parameters7 Crash severities Baseline padding 3 Paddings Real-life like
crash situationsRandomly selected
paddings Real-life like crash situations
Real-life like crash
situations
Baseline Pedal 2 Occupant sizes3 Foot airbags
Foot model evaluation 3 Foot positions Foot model evaluation 4 Pedals Randomly selected
airbags Hybrid III dummyFoot model
evaluation Evaluation of the safety performance of the car model
Evaluation of countermeasures in real-life crash situation models Human foot model
Fig. 1 - Design of the parametric studies
MODELS
In the baseline vehicle model, the seat, the bracket of the knee bolster and steering wheel and column
were multi-body. Toepan, carpet, frontal airbag, foot airbag and knee panel were finite elements.
Acceleration pulses and intrusion data of the toepan were obtained from full scale finite element
simulations with validated car structural models. The intrusion was applied as pre-scribed motion of
each node of the toepan and carpet. The toepan intrusion from the structural fem models were
converted to carpet intrusion by pre-simulations in which the carpet was substituted with a material
having a low elasticity modulus. Then the substituted carpet was forced to fall on the toepan. In the
end of the pre-simulations, the carpet was tight to the toepan and each node of the carpet was replaced
with the corresponding node number of the toepan. The relative intrusion of these nodes was used as
intrusion of the carpet, with start at the original carpet position. The baseline model was compared to
crash test data. The characteristics of the seat and the crush angle of the knee panel were similar as in
the baseline model in Höglund et al. (2002). The contact definition between the dummy and airbag
and between dummy and pedal were also similar as in the baseline model in Höglund et al. (2002).
The motion of the pedal in the baseline model was estimated from crash test films. The seat belts were
standard MADYMO belt elements with pretensioner.
Fig. 2 - The baseline model
(left) and the toepan (right)
STUDY 1, EVALUATION OF THE MODEL IN REAL-LIFE CRASH SITUATIONS
The evaluation of the protection ability of the car in different crash situation was done by
comparing the responses of different dummies, in different positions, in a number of crash situations
which are listed in Table 1. The impact velocities can not be published, but velocity B is higher than
velocity A and so on. The crash severities were selected to represent a broad spectrum of crash
situations with six different impact velocities, two barrier configurations and three impact directions.
Each crash severity got its unique intrusion of the toepan. The intrusions of a node below the brake
pedal for each crash situation are listed in Table 1. There were large differences in intrusion between
different crash situations. For example, the intrusion distance in crash #4 was 0.08 m and peak
acceleration 1685 m/s2. The peak acceleration in crash #5 was only 833 m/s2 for 0.11 m intrusion.
IRCOBI Conference – Lisbon (Portugal), September 2003 164
Table 1: Crash severities in the parametric studies
Impact Toepan intrusion
Number Reference Barrier Direction
Onset time
[ms]
Distance
[m] 1
Peak velocity
[m/s] 1
Peak acceleration
[m/s2] 1
1 Vel. A, FFRB Rigid Full frontal - 0.01 (62) 0.2 (58) 98 (21)
2 Vel. B, FFRB Rigid Full frontal 45 0.05 (59) 4.7 (50) 1005 (46)
3 Vel. B, 30°RB Rigid 30° 40 0.05 (74) 3.8 (58) 472 (55)
4 Vel. C, FFRB Rigid Full frontal 35 0.08 (69) 6.4 (39) 1685 (36)
5 Vel. D, ODB Deformable 40% offset 65 0.11 (101) 5.8 (77) 833 (66)
6 Vel. E, ODB Deformable 40% offset 59 0.17 (100) 10.0 (66) 1616 (61)
7 Vel. F, ODB Deformable 40% offset 49 0.29 (101) 11.8 (63) 1883 (51) 1 First number is peak and the second number is the time for peak in milliseconds
The standard MADYMO hybrid III dummies (TNO, 1999) were used in the simulations,
equipped with standard feet or biofidelic foot models. Two occupant sizes in three foot positions were
investigated (Table 2). The biofidelic lower leg, foot and ankle model was developed by
Hall et al. (1998) and improved by Dubbeldam et al. (1999). The biofidelic foot modelled the major
bones, joints and ligaments in the foot. The tendons were passive with loading functions according to
Hall et al. (1998) and Dubbeldam et al. (1999). Originally, the foot modelled only the right foot of a
50th percentile male without shoes. In this study, the model was mirrored to the left lower leg,
implemented in the Hybrid III and equipped with shoes, similar as in the MADYMO Hybrid350
dummy (TNO, 1999).
Table 2: Occupant size, model complexity and foot position in the parametric studies Parameter Reference Level
HIII, 50th M Hybrid III, 50th percentile male, standard foot model
HIII, 5th F Hybrid III, 5th percentile woman, standard foot model
Human, 50th M Hybrid III, 50th percentile male with biofidelic foot model
Occupant
model
Human, 5th F Hybrid III, 5th percentile woman with biofidelic foot model
Baseline Straight on the accelerator and on the foot rest, as in Euro-NCAP
Angled Both feet moved inward and the feet angled outward to the accelerator and foot rest Foot
position Brake pedal Right foot positioned on the brake pedal
The feet were scaled from 50th percentile male to 5th percentile woman using scale factors. Scale
factors for length in x, y and z direction was presented by Mertz et al. (1989). Mertz et al. (1989)
suggested the force ratio is based on the scale factor for dimension in x and z-direction, which ends up
with a force ratio of 0.68. However, a more recent study by Yoganandan et al. (2000) showed that the
scale factors for load and moment should be 0.63 and 0.5 respectively and these scale factors were
used in this study.
No scale factor for moment of inertia was found in the literature but a simple unit comparison
was conducted as approximation. The unit of moment of inertia is [kgm2] and thus λix, iy, iz=λmass⋅λ2x, y, z.
The shoes of the female dummy were similar as the Hybrid III, 5th percentile female shoes. The scaled
model was not compared with test data. The scale factors are listed in Table 3.
Table 3: Scale factors for the human foot model, 5th percentile female Property Scale factor Property Scale factor
λx 10.85 λmass
10.545
λy 10.85 λix; λiy; λiz 0.39; 0.39; 0.35
λz 10.80 λload
20.63
λmoment 20.5
1. Mertz et al. (1989)
2. Yoganandan et al. (2000).
STUDY 2A, EVALUATION OF COUNTERMEASURES, FIXED LEVELS
The objective of the simulations in this part of the study was to evaluate the model under real-life
safety conditions and investigate countermeasures; different characteristics of toepan padding, foot
airbag and pedal release (Figure 3). The airbag model was just a prototype and thus, all properties
related to the airbag design, such as gas inflation and vent hole are not relevant. However, a foot
IRCOBI Conference – Lisbon (Portugal), September 2003 165
airbag causes a motion of the carpet, which is independent of the airbag design. For that reason, the
bag was replaced with a finite element shell with a pre-scribed motion similar to the top of the bag.
That enabled studies of different airbag motions. This simplification reduced the simulation time
greatly. The airbag model was made of steel and deformed plastically in the simulations. The
simulation time was 6.5 hours. Without bag, but with a pre-motioned shape, the simulation time was
reduced to approximately 5 minutes for the Hybrid III, 50th male and 15 minutes for the Hybrid III,
5th female. Four different pedals were investigated as shown in Figure 3. The baseline pedal (pedal0)
did not release. Pedal1 released at 1.7 kN and descended towards the toepan. Pedal2 released and
moved linearly towards the toepan according to a load-displacement translational joint and pedal3
released early in the crash event.
0
0.02
0.04
0.06
0.08
0.1
0 0.02 0.04 0.06
Time [s]
Dis
pla
ce
me
nt
[m]
Footbag1Footbag2Footbag3
0E+0
2E+7
4E+7
6E+7
0 0.5
Strain
Str
ess [
N/m
2]
1
Padding0Padding1Padding2
0
2000
4000
6000
0.00 0.05 0.10
Displacement [m]
Fo
rce
[N
]
Pedal 0Pedal 1Pedal 2Pedal 3
Fig. 3 - Foot airbag (left), toepan padding (centre), pedal joint load characteristics (right)
All combinations of the parameters listed in Table 1 and Table 2 and shown in Figure 3 were
simulated.
STUDY 2B, EVALUATION OF COUNTERMEASURES, STOCHASTIC APPROACH
In order to get an improved understanding of the influence of foot bag characteristics and
padding material properties on the responses, 873 additional simulations were run using the baseline
model with the padding characteristics randomly selected. Another 181 simulations were conducted
with an added airbag and the foot airbag characteristics were stochastically chosen. The variables were
inflation onset time, motion of the top of the bag (simulates inflation volume) and end inflation time.
The crash situation and foot position were selected according to a random function in both
simulation series. All randomly selected variables were evenly distributed. In this part of the study, all
simulations were conducted with the Hybrid III 50th percentile male dummy with standard foot model.
That enable simulations with angled foot and the CPU time was significantly reduced.
RESULTS
The first 5 ms of the response-history curves were removed to leave out initial effects. The
simulations with the human foot initially angled indicated that the model was not realistic in such
situations and these results are not presented in Figure 4 or appendix 2-5 and accordingly, no
conclusions are drawn from these results. Several simulations with the human foot model, especially
for crash #7, were unstable due to numerical errors. In general, the difference in axial tibial load
between simulations with the human-like foot model and the Hybrid III dummy foot were smaller than
the differences in ankle and foot rotation.
STUDY 1, EVALUATION OF THE MODEL IN REAL-LIFE CRASH SITUATIONS
Figure 4 shows that the right axial tibial force in Hybrid III and the human foot model were
almost equally influenced of crash situations for both dummy sizes. The right axial tibial force was in
general lowest for crash #5, an offset crash, and crash #3, an angled impact. The highest forces
occurred for crash #4, frontal crash.
The male dummy, both with the standard foot model and the human foot model was least
influenced of the crash situation when the foot was initially positioned on the brake pedal. The male
dummy was least influenced of the foot position in crash #5, both with the human foot model and the
standard Hybrid III foot model.
IRCOBI Conference – Lisbon (Portugal), September 2003 166
The female dummy varied more than the male dummy between different crash situations, foot
positions and dummy foot biofidelity. For example, there was a rather large difference in right tibia
load between the female dummy with the foot initially angled on the accelerator than in the baseline
position for the most severe offset crash, crash #7. The foot slipped off the pedal and hit the toepan at
a time with high toepan intrusion velocity. The axial load in the right tibia in the female dummy
increased with impact velocity when the right foot was positioned on the brake pedal.
The left axial tibial forces were in general lower than the right tibial forces (Appendix 2). The
differences in axial force across the crash situations turned out to be relatively small for the Hybrid III,
50th percentile male. The left axial tibial force in the Hybrid III, 5th percentile female with the foot
initially angled, varied considerable between the crash conditions. With the left foot angled, the female
dummy had higher responses for crashes with angled impact direction or offset impact than full frontal
impact, i.e. higher force with a lateral motion than without.
0
1000
2000
3000
4000
Vel. A, FFRB Vel. B, FFRB Vel. B, 30°RB Vel. C, FFRB Vel. D, ODB Vel. E, ODB Vel. F, ODB
Axia
l fo
rce
[N
]
HybridIII, 5th Female, Baseline HybridIII, 5th Female, Angled HybridIII, 5th Female, Brakepedal
HumanIII, 5th Female, Baseline HumanIII, 5th Female, Brakepedal
0
1000
2000
3000
4000
5000
6000
7000
Vel. A, FFRB Vel. B, FFRB Vel. B, 30°RB Vel. C, FFRB Vel. D, ODB Vel. E, ODB Vel. F, ODB
Axia
l fo
rce
[N
]
HybridIII, 50th Male, Baseline HybridIII, 50th Male, Angled HybridIII, 50th Male, Brakepedal
HumanIII, 50th Male, Baseline HumanIII, 50th Male, Brakepedal
Fig. 4 - Axial force in right tibia for different foot positions, dummy foot and crash severity for
5th percentile female dummy (top) and 50th percentile male dummy (bottom)
Almost all dummy models dorsiflexed as most in the right ankle when the foot was positioned on
the brake pedal. The peak dorsiflexion angle was significantly higher in the Hybrid III foot than the
human foot (Appendix 3). The dorsiflexion angle in the human foot model was never more than 37°.
The human foot was also very stiff in the tarsal joint and everted/inverted just a few degrees
(Appendix 4).
STUDY 2, EVALUATION OF COUNTERMEASURES
Three different countermeasures were evaluated; toepan padding, foot airbag and pedal
deformation characteristics. The difference in axial force in right tibia between padding0 (baseline
padding), padding1 and padding2 was small (Appendix 5). However, the simulations with randomly
selected padding properties showed that it was possible to get lower axial force in this crash situation
using other padding materials. The lines in Figure 5 show the interval between maximum and
minimum of the simulations with randomly selected padding characteristics. The lowest force in
crash #4 was 4600 N for the most optimum padding property. Of the investigated padding materials,
the highest force in crash #4 was 6800 N. For the crash situation #6, the difference between highest
force and lowest force was 2600 N and for crash #5, the difference just 500 N.
IRCOBI Conference – Lisbon (Portugal), September 2003 167
0
2000
4000
6000
8000
Vel. A,
FFRB
Vel. B,
FFRB
Vel. B,
30°RB
Vel. C,
FFRB
Vel. D,
ODB
Vel. E,
ODB
Vel. F,
ODB
Axia
l fo
rce
[N
]
Fig. 5 - The lines show the interval between
maximum and minimum of the simulations
with randomly selected padding
characteristics (onset inflation time,
distance, end inflation time). The bars
represent right tibia axial load with baseline
pedal, padding and foot position.
The airbag in the foot area was represented by a motion of the carpet. Figure 6 shows the average
axial force in right tibia for the Hybrid III, 50th percentile male and 5th percentile female, in different
foot positions for different foot airbags. Bag #1 and #2 reduced the axial force in the Hybrid III,
50th percentile male in baseline position. When the foot was angled on the accelerator or initially
positioned on the brake pedal, the force was slightly higher with foot airbag. The axial force in the
female dummy was uninfluenced of bag 1 in the baseline foot position, but the force was substantial
higher in the combination of angled position and foot airbag.
0
2000
4000
6000
305,
Baseline
305,
Angled
305,
Brake
pedal
350,
Baseline
350,
Angled
350,
Brake
pedal
Axia
l fo
rce
[N
]
Bag1 Bag2 Bag3 Baseline countermeasure
Fig. 6 - Average axial force in
right tibia for the Hybrid III,
50th percentile male and
5th percentile female in different
foot positions for different foot
airbags. The dotted bars
represent simulations with the
same countermeasures as in the
baseline model.
The deviation in forces among the simulations with randomly selected foot airbag characteristics
was large (Appendix 6). Figure 7 shows the average axial force in right tibia for randomly selected
foot bag properties in the model with baseline pedal, padding and foot position. High axial forces
occurred in the right tibia in crash situation #4. The load in tibia was 2900 N with an optimized foot
bag, which was a reduction with 50% compared to the simulations with baseline padding.
0
2000
4000
6000
8000
10000
12000
Vel. A,
FFRB
Vel. B,
FFRB
Vel. B,
30°RB
Vel. C,
FFRB
Vel. D,
ODB
Vel. E,
ODB
Vel. F,
ODB
Axia
l fo
rce
[N
]
Fig. 7 - The lines show the interval between
maximum and minimum of the simulations
with randomly selected foot bag properties
(onset inflation time, distance, end inflation
time). The bars represent right tibia axial
load with baseline pedal, padding and foot
position.
The difference in tibia load for different pedals was small. However, the average right ankle
rotation and right foot rotation were slightly reduced with pedal 3, a pedal that deforms early in the
crash event (Figure 8). This was especially obvious for right ankle eversion for the 50th percentile male
with the foot angled on the accelerator.
IRCOBI Conference – Lisbon (Portugal), September 2003 168
0
10
20
30
40
Pedal0 Pedal1 Pedal2 Pedal3
Do
rsifle
xio
n a
ng
le [
de
g]
HybridIII, 5th F, BaselineHybridIII, 5th F, AngledHybridIII, 50th M, BaselineHybridIII, 50th M, AngledHumanIII, 5th F, BaselineHumanIII, 50th M, Baseline
-40
-30
-20
-10
0
10
20
30
40
50
Pedal0 Pedal1 Pedal2 Pedal3Eve
rsio
n (
-)/I
nve
rsio
n (
+)
[de
g] HybridIII, 5th F, Baseline
HybridIII, 5th F, AngledHybridIII, 50th M, BaselineHybridIII, 50th M, AngledHumanIII, 5th F, BaselineHumanIII, 50th M, Baseline
Fig. 8 - Average right dorsiflexion angle (left) and eversion/inversion (right) for different pedal
characteristics. Pedal0 is the baseline pedal.
DISCUSSION
An evaluation of the safety performance of a car in real-life crash situations has been performed
with mathematical dummy-vehicle models. A number of simulation series, in total almost 3200 runs,
were made using the models to evaluate the toepan design, safety systems and the dummy model in
different crash situations. The vehicle model was designed and compared to a number of crash tests.
Pelvis acceleration, tibial axial forces and bending moments showed acceptable agreement with crash
tests (Appendix 1). The femur forces agreed well in magnitude but the duration of the femur forces in
the model was shorter than the compared crash test dummy responses.
Previous studies by the authors have indicated that the responses depend on a large number of
parameters describing the toepan configuration. Some parameters influenced the responses differently
depending of the levels of other parameters and these interaction effects have to be taken into account
in lower limb protection evaluation (Höglund et al., 2003). This requires a large number of
simulations, and to reduce the CPU-time, the models have to consist of as few finite element parts as
possible. Most parts in the model were multi-body systems. The toepan, knee bolster, carpet, foot
airbag and frontal airbag were modelled as finite elements to get a shape during intrusion as similar as
possible to real life crash situations.
The model did not include a frontal airbag in the parametric studies. A brief study was conducted
to evaluate the influence of the frontal airbag. 28 combinations with human foot were selected
randomly from the test matrix. The models were simulated both with and without airbag. The
percentage difference with and without airbag were calculated for each lower limb response in all
combinations. The average difference between all peak responses was 1.5%. However, for the most
interesting responses, peak axial tibial forces, foot and ankle rotations and moments, the average was
less than 1%. These results correlate well with results by Crandall et al. (1997) who demonstrated that
airbags have not reduced the frequency of lower leg injuries. Since the difference in responses with
and without frontal airbag was small, the parametric studies were conducted without frontal airbag to
reduce the CPU time.
An important advantage with this study was the use of a biofidelic dummy foot model
(Hall et al., 1998, Dubbeldam et al., 1999). The model consisted of the most important bones, joints
and ligaments of the foot. Unfortunately, the knee joint was similar as in the Hybrid III dummy, a
combination of a translational joint and a revolute joint. Therefore, the external and internal rotation of
the lower leg could not be done in the knee-tibia-fibula-complex. These rotations took place in the
tarsal joints and therefore the ligaments pulled the foot back to the original position. The simulations
with the human like foot model angled on the accelerator and foot rest was then not as realistic as the
simulations with the feet in the baseline or brake pedal position. The results from simulations with the
human foot initially angled are therefore excluded.
The ligaments were modelled as belt elements and Maxwell elements. In some crash situations,
numerical errors occurred when any ligaments become zero in length. This problem occurred most
often for the short ligaments between medial and lateral malleolus and talus. The ligaments in the foot
IRCOBI Conference – Lisbon (Portugal), September 2003 169
strongly influenced the kinematics of the biofidelic foot model and the difference compared to the
Hybrid III foot model was particularly in foot rotation, ankle rotation and bending moments.
The axial force was measured in the same position, distal tibia, in both the standard Hybrid III
foot model and the human-like foot model, which reduced the influence of the difference in tibia shape
between the models. Despite differences in kinematics, peak axial forces in especially the
50th percentile male correlated fairly well in the brake pedal and baseline position. The axial loads
were in general higher in the human foot model than the Hybrid III model, which was due to the
Achilles tendon. To investigate the influence of the Achilles tendon, all crash situations with baseline
countermeasures and with the human-like foot model were compared with simulations without an
Achilles tendon. These simulations indicated that the Achilles tendon was the main source for the
difference between the human-like models and the Hybrid III models. For situations with the foot on
the pedal during the whole crash event, there were just small differences in axial load between the
Hybrid III dummy foot and the human like foot model. However, simulations with the female dummy
showed that the ankle rotation often were small, which caused small differences in axial load between
the models with and without a passive Achilles tendon.
The largest difference between the male foot models in baseline position was at low impact
velocities, crash #1 and #2, and in brake pedal position, at angled impact. In that situation, the peak
force occurred at impact to the toepan and in the human foot model, this correlated with a rather high
load in the Achilles tendon.
The axial tibial load in the female dummy with Hybrid III standard shoe was higher than the
dummy with the biofidelic foot model in crash #4. The differences in kinematics were small in that
crash situation, but large enough for the Hybrid III foot to impinge more in the padding than the
human-like foot.
Funk et al. (2002) found that malleolar fractures were common in eversion for ankles preloaded
with 2 kN. In tests with similar conditions and the ankles also initially dorsiflexed 30° resulted in talar
fractures in their study. Therefore, even though axial forces match fairly well, correct description of
the rotation of both the feet and the ankle are important in studies of the interaction between occupants
and vehicle for comparing dummy responses with real-life injuries. It would thus be interesting to
improve the human foot model to be stable in most crash conditions and with a biofidelic
external/internal rotation.
The evaluation of the car model in real-life crash situations showed that the responses were
highly dependent of the crash situation, and within a particular crash with a reasonable variation in
parameters defining the safety systems and occupant. The toepan intrusion was different in all crash
situations and there were no distinct correlation between any response and any parameter describing
the intrusion. With the baseline model and baseline position, the right ankle dorsiflexion in the
Hybrid III, 50th percentile male dummy correlated with intrusion acceleration following a equation of
the second degree (R2=0.91), but this was not the case for any other dummies or foot position. In some
simulations, the axial load was lower when the foot was positioned on the brake pedal, which
correlates with findings in a simulations study be Pilkey et al. (1994). In the baseline position,
50th percentile male dummy and baseline countermeasures, there were a correlation between peak
acceleration of the toepan and tibia load; the load were highest in the three crashes with highest toepan
acceleration. This correlates well with studies by Kuppa and Sieveka (1995) and Sakurai (1996)
Three safety systems were evaluated; foot airbags, toepan paddings and deformable pedals. The
examination of the influence of toepan padding indicated that the property of the padding significantly
influenced the axial load at crashes with high peak acceleration. Crashes with low peak acceleration,
as crash #5 were just slightly influenced of different padding properties. No correlation was found
between intrusion magnitude and the influence of different toepan paddings on tibia load.
The baseline motion of the foot airbag was given by the average node motion of a simulation
with a finite element foot bag. The results from the study indicated that a foot airbag have the potential
to reduce tibial forces. A foot airbag with onset time 12.2 ms, end inflation time 19.6 ms and total
motion, or thickness at full inflation, of 0.0795 m could be effective, especially in combination with
pedal #3. In order to evaluate these countermeasures in different crash situations, foot positions and
dummy sizes, a two-level factorial design study was done (Box et al., 1978). The effects showed that
the most important effects were dummy size, toepan configuration and the interaction between these
factors. The interaction in axial force between dummy and toepan configuration for each crash
IRCOBI Conference – Lisbon (Portugal), September 2003 170
situation are summarized in Table 4. As shown in the table, a foot airbag reduced the axial force in
right tibia for the 50th percentile male, baseline positioned, in most situations. For the 5th percentile
female, a foot bag increased the force significantly in four crash situations. However, when the right
foot was positioned on the brake pedal, a foot bag reduced the axial force in the female dummy but not
in the larger male dummy. A foot airbag optimized for a severe crash situation as full frontal in rather
large impact velocity got the potential to reduce axial loads in several crash situations. However, the
bag increased the tibial load in 50% of the simulations in the two-level analysis.
Table 4: Average right axial force in the two-level factorial analysis
Position Dummy
Counter-
measure
Vel. A,
FFRB
Vel. B,
FFRB
Vel. B,
30°RB
Vel. C,
FFRB
Vel. D,
ODB
Vel. E,
ODB
Vel. F,
ODB
Baseline HIII, 5th F Baseline 1383 89 1792 3885 469 1122 1844
Baseline HIII, 5th F Foot bag 1130 1051 1195 3055 2041 3349 3339
Baseline HIII, 50th M Baseline 4735 2888 5149 6065 2489 4792 5819
Baseline HIII, 50th M Foot bag 2112 2079 2257 2836 2800 4547 5962
Brake HIII, 5th F Baseline 405 853 485 1447 1533 3277 3342
Brake HIII, 5th F Foot bag 1085 537 1049 1122 2109 1719 2052
Brake HIII, 50th M Baseline 3008 1548 2792 2873 2588 3927 4311
Brake HIII, 50th M Foot bag 2067 2524 2579 3459 2958 4532 5417
The results in this study correlates well with findings in Hesse et al. (2001), who found in a series
of sled tests that a foot airbag reduced the axial force. However, this study demonstrates that the tibial
axial force become higher with bag than without bag in 50% of all simulations with a foot bag in the
two-level factorial analysis. Introducing a foot airbag is though the only method of the
countermeasures investigated in this study to get right tibia axial forces below 4500 N in a full frontal
crash, rigid barrier, as crash situation #4 keeping the similar toepan intrusion and foot positions. The
foot airbag have to be optimized for a number of crash situations and it would be advantageous to be
able to adjust the airbag inflation depending on the crash situation, driver stature and foot position.
CONCLUSIONS
The evaluation of the mathematical vehicle model in a number of crash situations showed that the
dummy responses varied significantly for different crash situations. Therefore, implementing foot and
ankle safety systems in cars have to be preceded by an investigation of its effects on occupant
responses in a broad spectrum of crash situations, occupant sizes and foot positions. The simulations
with a biofidelic dummy foot showed that the dorsiflexion and inversion/eversion of the ankle and foot
differed significantly compared to the Hybrid III foot, but the axial forces matched fairly well.
The pedal was not a major source for high axial tibial forces in this study, but the eversion of the
right foot was lower when the pedal deformed early in the crash event.
A foot airbag reduced axial tibial loads in some crash situations, but the simulations indicated
that the bag had to be optimized for a number of crash situations. This study showed that an airbag
also could increase the responses for some combination of occupant size, crash situation and foot
position. The axial tibial load was higher in fifty percent of all simulations with a foot airbag
compared to similar crash situation without airbag in this study.
None of the studied padding materials could reduce tibial forces below 4500 N in a full frontal
crash with rigid barrier at velocity C.
ACKNOWLEDGEMENTS
This work was funded by Saab Automobile AB and Programrådet för Fordonsteknisk Forskning
(PFF). The authors would like to thank Mrs Rosemary Dubbeldam at Delphi Automotive Systems and
Prof. Jeff R. Crandall at the Center for Applied Biomechanics at University of Virginia who developed
and provided the biofidelic foot model.
IRCOBI Conference – Lisbon (Portugal), September 2003 171
REFERENCES
Begeman, Paul; Balakrishnan, Pradeep; Levine, Robert; King, Albert I., ‘Dynamic human ankle
response to inversion and eversion’ 37th Stapp Car Crash Conference, San Antonio, Texas, USA,
1993, pp 83-93
Box, G. E. P.; Hunter, W. G.; Hunter, S. J. ‘Statistics for Experimenters’ 1978 John Wiley & Sons, Inc.
ISBN 0-471-09315-7.
Crandall, Jeff R.; Martin, Peter G.; Bass, Cameron R.; Pilkey, Walter D.; Dischinger, Patricia C.;
Burgess, Andrew R.; O'Quinn, Timothy D.; Schmidhauser, Carl B., ‘Foot and ankle injury: the
roles of driver anthropometry, footwear and pedal controls’ 40th Annual Proceedings of the
Association for the Advancement of Automotive Medicine, Vancouver, British Columbia, Canada,
1996 pp 1-18
Crandall, J. R.; Martin, P. G, ‘Lower limb injuries sustained in crashes and corresponding
biomechanical research’ International Symposium on Real World Crash Injury Research,
Leicestershire, England, 1997 pp 3.5.1-3.5.16
Crandall, Jeff R.; Martin, Peter G.; Sieveka, Edwin M.; Pilkey, Walter D.; Dischinger, Patricia C.;
Burgess, Andrew R.; O'Quinn, Timothy D.; Schmidhauser, Carl B, ‘Lower limb response and
injury in frontal crashes’, Accident Analysis and Prevention v30 n5 Sep 1998 p 667-677 0001-
4575 AAPVB5
Dischinger, P. C.; Kufera, J. A.; Kerns, T. J. ’Lower extremity fractures in motor vehicle collisions:
The role of driver gender and height’, 38th Annual Proceedings Association for the Advancement
of Automotive Medicine, Lyon, France, 1994 pp 601-605
Dubbeldam, R.; Nilson, G.; Pal, B.; Eriksson, N.; Owen, C.; Roberts, A.; Crandall, J.; Hall, G.;
Manning, P.; Wallace, A., ‘A MADYMO model of the foot and leg for local impacts’ 43rd
Annual Stapp Car Crash Conference, San Diego, California, USA, 1999, pp 185-202
Fildes, B.; Lenard, J.; Lane, J.; Seyer, K., Lower limb injury in frontal crashes. International
conference on pelvic and lower extremity injuries, Washington, DC USA, 1995 pp 1-10
Funk, J. R.; Srinivasan, S.C. M.; Crandall, J. R. ; Khaewpong, N; Eppinger, R. H.; Jaffredo, A. S.; ;
Petit, P. Y., ‘The effects of axial preload and dorsiflexion on the tolerance of the ankle/subtalar
joint to dynamic inversion and eversion’, 46th Stapp Car Crash Conference, Ponte Verdra Beach,
Florida, USA 2002, pp 245-265
Hall, G. W.; Crandall, J. R.; Pilkey, W. D.; Thunnissen, J. G., ‘Development of a dynamic multibody
model to analyze human lower extremity impact response and injury’ International IRCOBI
Conference on the Biomechanics of Impact, Göteborg, Sweden, 1998 pp 117-134
Hesse, S.; Bläßer, S.; Hoffmann, J., ‘Parameter study on different factors influencing lower extremity
injuries’ 17th International Technical Conference on the Enhanced Safety of Vehicles Amsterdam,
The Netherlands, 2001 Paper #101
Håland, Y.; Hjerpe, E.; Lövsund, P., ‘An inflatable carpet to reduce the loading of the lower
extremities - evaluation by a new sled test method with toepan intrusion’ 16th International
Technical Conference on the Enhanced Safety of Vehicles, Windsor, Ontario, Canada, 1998 pp
292-301
Höglund, N.; Lövsund, P.; Viano, D.C. ‘Parametric investigation of the toepan region using
mathematical models’ International Journal of Crashworthiness 2002 (Submitted)
Höglund, N.; Lövsund, P.; Viano, D.C. ‘Mathematical modelling of lower leg responses in offset
crashes’ Journal of Crashworthiness 2003 Vol 8 No 3, pp 285-295
Kippelt, Ulrich; Buss, Winfried; Feldhoff, Uwe; Thelen, Martin, ‘Protection devices and development
tools for reducing foot and leg injuries in frontal crashes’ International IRCOBI Conference on
the Biomechanics of Impact, Göteborg, Sweden, 1998 pp 161-172
IRCOBI Conference – Lisbon (Portugal), September 2003 172
Kuppa, S. M.; Sieveka, E. M., 'Dynamic motion of the floor pan and axial loading through the feet in
frontal crash tests', International IRCOBI Conference on the Biomechanics of Impact, Brunnen,
Switzerland, 1995, pp 389-402
Kuppa, S.; Wang, J.; Haffner, M.; Eppinger, R., ’Lower extremity injuries and associated injury
criteria’ International Technical Conference on the Enhanced Safety of Vehicles, Amsterdam,
The Netherlands, 2001, Paper #457
MacKenzie EJ, Cushing BM, Jurkovich GJ, Morris JA Jr, Burgess AR, deLateur BJ, McAndrew MP,
Swiontkowski MF. ‘Physical impairment and functional outcomes six months after severe lower
extremity fractures’ J Trauma. 1993 Apr;34(4):528-38
Manning, P.; Owen, C.; Roberts, A.; Oakley, C.; Lowne, R., ‘Dynamic Response And Injury
Mechanism In The Human Foot And Ankle And An Analysis Of Dummy Biofidelity’, 16th
International Technical Conference on the Enhanced Safety of Vehicles, Windsor, Ontario,
Canada, 1998, pp 1960-1998
Mertz, H. J.; Irwin , A. L.; Melvin, J. W.; Stalnaker, R. L.; Beebe, M. S.; ‘Size, weight and
biomechanical impact response requirements for adult size small female and large male
dummies’ SAE International Congress and Exposition, Detroit, Michigan, USA 1989, pp 133-
140
Pilkey, Walter D.; Sieveka, Edwin M.; Crandall, Jeff R.; Klopp, Gregory, ‘The influence of foot
placement and vehicular intrusion on occupant lower limb injury in full-frontal and frontal-offset
crashes’ 14th International Technical Conference on the Enhanced Safety of Vehicles, Munich,
Germany, 1994, pp 734-741
Richter, M.; Thermann, H.; Wippermann, B.; Otte, D.; Schratt, HE.; Tscherne, H., ‘Foot fractures in
restrained front seat car occupants: a long-term study over twenty-three years’, J Orthop Trauma
2001 May;15(4):287-293
Sakurai, M. 'An analysis of injury mechanisms for ankle/foot region in frontal offset collisions',
40th Stapp Car Crash Conference, Albuquerque, New Mexico, USA, 1996
Thomas, P; Charles, J.; Fay, P., ‘Lower limb injuries~The effect of intrusion, crash severity and the
pedals on injury risk and injury type in frontal collisions’ 39th Stapp Car Crash Conference, San
Diego, California, USA, 1995 pp 265-280
TNO (1999) ‘MADYMO User’s Manual 3D’ TNO Road-Vehicles Research Institute, Delft The
Netherlands
Yoganandan, N; Pintar, F. A.; Pintar, F. A.; Kumaresan, S; Kumaresan, S; Gennarelli, T; Gennarelli,
T. A.; Sun, E; Kuppa, S; Maltese, M; Eppinger, R. H., ’Pediatric and small female neck injury
scale factors and tolerance based on human spine biomechanical characteristics’, International
IRCOBI Conference on The Biomechanics of Impact, Montpellier, France 2000
IRCOBI Conference – Lisbon (Portugal), September 2003 173
APPENDIX
Gray shaded lines represents crash test 1, 2 and 3. Black line is the baseline MADYMO model
-700
-600
-500
-400
-300
-200
-100
0
100
0 25 50 75 100 125
Time [ms]
Pelv
is x
-acc [
m/s
2]
-500
-400
-300
-200
-100
0
100
200
0 25 50 75 100 125
Time [ms]
Pelv
is z
-acc [
m/s
2]
-2000
-1000
0
1000
2000
3000
4000
0 25 50 75 100 125
Time [ms]
Rig
ht
fem
ur
forc
e [
N]
-2000
-1000
0
1000
2000
3000
4000
5000
0 25 50 75 100 125
Time [ms]
Left
fem
ur
forc
e [
N]
-1000
0
1000
2000
3000
4000
5000
6000
0 25 50 75 100 125Time [ms]
Rig
ht
tib
ia f
orc
e [
N]
-1000
0
1000
2000
3000
4000
5000
6000
0 25 50 75 100 125Time [ms]
Left
tib
ia f
orc
e [
N]
-100
-50
0
50
100
150
0 25 50 75 100 125
Time [ms]
Lw
r ri
gh
t tib
ia M
y
[Nm
]
-80
-60
-40
-20
0
20
40
0 25 50 75 100 125
Time [ms]
Lw
r le
ft t
ibia
My
[N
m]
Appendix 1: Crash test and
MADYMO simulations with the
baseline models in full frontal rigid
barrier crash situation
0
1000
2000
3000
4000
5000
Vel. A, FFRB Vel. B, FFRB Vel. B,
30°RB
Vel. C, FFRB Vel. D, ODB Vel. E, ODB Vel. F, ODB
Axia
l fo
rce
[N
]
HybridIII, 5th F, Baseline HybridIII, 5th F, Angled HybridIII, 50th M, Baseline
HybridIII, 50th M, Angled HumanIII, 5th F, Baseline HumanIII, 50th M, Baseline
Appendix 2: Axial force in left tibia for different dummies and foot positions
0
10
20
30
40
50
60
Vel. A, FFRB Vel. B, FFRB Vel. B, 30°RB Vel. C, FFRB Vel. D, ODB Vel. E, ODB Vel. F, ODB
Dors
ifle
xio
n [
deg]
HybridIII, 5th F, Baseline HybridIII, 5th F, Angled HybridIII, 5th F, BrakepedalHybridIII, 50th M, Baseline HybridIII, 50th M, Angled HybridIII, 50th M, BrakepedalHumanIII, 5th F, Baseline HumanIII, 5th F, Brakepedal HumanIII, 50th M, BaselineHumanIII, 50th M, Brakepedal
Appendix 3: Right ankle rotation, dorsiflexion, for different dummies and foot positions
IRCOBI Conference – Lisbon (Portugal), September 2003 174
-60
-40
-20
0
20
40
60
Vel. A, FFRB Vel. B, FFRB Vel. B, 30°RB Vel. C, FFRB Vel. D, ODB Vel. E, ODB Vel. F, ODBEve
rsio
n (
-),
Inve
rsio
n (
+)
[de
g]
HybridIII, 5th F, Baseline HybridIII, 5th F, Angled HybridIII, 5th F, BrakepedalHybridIII, 50th M, Baseline HybridIII, 50th M, Angled HybridIII, 50th M, BrakepedalHumanIII, 5th F, Baseline HumanIII, 5th F, Brakepedal HumanIII, 50th M, BaselineHumanIII, 50th M, Brakepedal
Appendix 4: Right foot rotation, inversion/eversion, for different dummies and foot positions
0
1000
2000
3000
4000
5000
6000
7000
Padding0 Padding1 Padding2
Axia
l fo
rce
[N
]
HybridIII, 5th F, Baseline HybridIII, 5th F, Angled HybridIII, 5th F, BrakepedalHybridIII, 50th M, Baseline HybridIII, 50th M, Angled HybridIII, 50th M, BrakepedalHumanIII, 5th F, Baseline HumanIII, 5th F, Brakepedal HumanIII, 50th M, BaselineHumanIII, 50th M, Brakepedal
Appendix 5: Average axial force in lower right tibia for different dummies, foot positions and toepan
padding materials.
0
2000
4000
6000
8000
10000
0 0.05 0.1 0.15Airbag displacement [m]
Axia
l fo
rce
[N
]
Appendix 6: Axial force in right tibia vs. airbag
displacement for all investigated airbags.
IRCOBI Conference – Lisbon (Portugal), September 2003 175